WTMaxMiner: Efficient mining of Maximal Frequent Patterns based on Weighted Directed Graph Traversals
Runian Geng, Xiangjun Dong, Ping Zhang, Wenbo Xu · 2008
Frequent itemset mining for traversal patterns have been found useful in several applications. However, (closed) frequent mining can generate huge and redundant patterns, and traditional model of traversal patterns mining considered only un-weighted traversals. In this paper, a transformable model between EWDG (Edge-Weighted Directed Graph) and VWDG (Vertex-Weighted Directed Graph) is proposed. Based on the model, an effective algorithm, called WTMaxMiner (Weighted Traversals-based Maximal Frequent Patterns Miner), is developed to discover maximal weighted frequent patterns from weighted traversals on directed graph. Experimental comparison results with previous work on synthetic data show that the algorithm has a good performance and scalable property to the problem of mining maximal frequent patterns based on weighted graph traversals.